midtropospheric co concentration derived from infrared and...

ITSC XVIAngra dos Reis, 7-13 May 2008

Cyril Crevoisier, Alain Chédin, Noëlle A. Scott, Gaelle Dufour, Raymond Armante and Virginie Capelle

Midtropospheric CO2 Concentration derived from infrared and microwave sounders.

Application to the TOVS, AIRS/AMSU, and IASI/AMSU instruments.

Laboratoire de Météorologie Dynamique, CNRS, IPSL, Palaiseau

CO2 from infrared/microwave sounders

Since 2006

IASI/AMSUESA/MetOp

AIRS/AMSUNASA/AquaSince 2002

NOAA10TOVS

NOAAkATOVS

1987-1991 1999-2005

19/15

7.30

19/4

7.30

AquaAIRS/AMSU

MetOpIASI/AMSU

Time coverage May 2002-… Oct. 2006-…

Spectral resolution

0.5 - 2 cm-1 0.5 cm-1

(apodized)# IR/MWchannels

2378/15(324/15)

8461/15(421/15)

Local time 1.30 9.30

NOAA10/HIRS/MSU1987-1991

NOAAk/HIRS/AMSU1999-2005

CO2 seasonal cycle - Northern tropics [0-20°N]

NOAA10TOVS

NOAA15ATOVS*

AquaAIRS/AMSU

MetOpIASI/AMSU

87 91 01 03 0793 95 97 99 0589340350360370380

•NOAA10: Chédin et al., JGR, 2003, 2008.

•AIRS: Crevoisier et al., GRL, 2004.

•NOAA15: Very preliminary results…

CO2 seasonal cycle - Northern tropics [0-20°N]

NOAA10TOVS

NOAA15ATOVS*

AquaAIRS/AMSU

MetOpIASI/AMSU

87 91 01 03 0793 95 97 99 0589340350360370380

Difference between7.30 am/pm observations

Diurnal Tropospheric Excess of CO2due to biomass burning emissions

See Poster B12 [Chédin et al.]

Spectrum and sensitivity to atmospheric components

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

CO2 (1%)H2O (20%)O3 (20%)CH4 (20%)CO (40%)N2O (2%)Surface

Sensitivity of IASI TB to variations of atmospheric and surface variables (simulations with the 4A RT model)

T Bpe

rt -T B

ref(K

)

10 μm 6 μm 4 μm15 μm

wavenumber (cm-1)

CO2

CO

CH4

surface

H2O

O3

N2O

1 % of CO2 variation → 0.04% of TB variation3ppmv → 0.3 K

The full information contained in the channels is needed to extract the CO2 signal!

Spectrum and sensitivity to atmospheric components

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

CO2 (1%)H2O (20%)O3 (20%)CH4 (20%)CO (40%)N2O (2%)Surface

Sensitivity of IASI TB to variations of atmospheric and surface variables (simulations with the 4A RT model)

T Bpe

rt -T B

ref(K

)

10 μm 6 μm 4 μm15 μm

wavenumber (cm-1)

CO2

CO

CH4

surface

H2O

O3

N2O

1 % of CO2 variation → 0.04% of TB variation3ppmv → 0.3 K

The full information contained in the channels is needed to extract the CO2 signal!

Spectrum and sensitivity to atmospheric components

CO2 (1%)H2O (20%)O3 (20%) Surf. emissivity (5%)

CO (40%)Surface T (1%)

Sensitivity of AIRS and IASI channels in the two CO2 bands(simulations with the 4A RT model)

AIRS

wavenumber (cm-1)

CO2

CO noiseH2OO3CO2noise

1.51

0.50

-0.5

1.51

0.50

T Bpe

rt -T B

ref(K

)-0.5

670 690 710 730 750 2220 2260 2300 2340 2380

IASI

4 μm15 μm

650

surface T

At 4 μm: Non-LTE

Only nightime retrieval

T JacobianCO2 Jacobian

O3 Jacobian

Jacobians of two “CO2” AIRS channels

Channel 80 - 15μm Channel 261 - 4μm

Spectrum and sensitivity to atmospheric components

10

100

1000

1

0.1

•Channels at 4μm peak lower in the atmosphere.

Jacobians of two “CO2” AIRS channelsand AMSU weighting functions

Channel 80 - 15μm Channel 264 - 4μm

AMSU 7AMSU 6

AMSU 8

AMSU 10

Spectrum and sensitivity to atmospheric components

T JacobianCO2 Jacobian

O3 Jacobian

10

100

1000

1

0.1

AMSU 9

•Channels at 4μm peak lower in the atmosphere.

•AMSU channels bring the information on temperature.

CO2 channel selection•Aqua AIRS/AMSU:

-15 AIRS channels (8 in the 15 μm band - 7 in the 4 μm band).-2 AMSU channels (6 and 8).

•MetOp IASI/AMSU:-14 IASI channels (all in the 15 μm band).-3 AMSU channels (6, 7, and 8).

IASI - CO2 channel selection

-0.05

0

0.05

0.1

0.15

0.2

0.25

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Sen

siti

vit

y (

K) CH4 4%

CO 20%

N2O 1%

O3 20%

CO2 1%

H2O 20%

AMSU 6

AMSU 8

AMSU 7

IASI CO2 JacobiansIASI sensitivity (K)

CO2 (1%) H2O (20%) O3 (20%) 10

100

1000

0.05

0

-0.05

0.1

0.15

0.2

0.25

The level1b validation suite at LMD

ECMWF

FTP « Clean »Radio-

soundings

L1B Satellite

Data(1979-2006)

RT model(Stransac, 4A)

Fixed CO2=372 ppmv

See Poster B10 [Armante et al.]

Radiative biasesbetween calculated and observed BT

Colocation of radiosoundings/re-analyses ERA40 with IR/MW observations

Space and Time

Colocation(100 Km, 3h)

Quality control of the Radiosoundings

Inter/extrapolationUGAMP climatology

Radio-soundings

/Reanalysis(ERA-40)

- Radiosoundings «ERA40 »(23 Go from 1979 to 2007 )-Re-analyses ERA-40 (79 Mo / day, 2 days /month)

Example IASI/AMSU (MetOp)Satellite data :14 orbits/day 900 Mo/day; 421 IR, 15 MW

AIRS Radiative biases: Monthly evolution (5 years over the tropics)

~60 situations/month over sea~30 situations/month over land

03 04 05 06 07 08-0.2

0

0.2

0.4

0.6

0.8

03 04 05 06 07 08

03 04 05 06 07 08

0.6

0.8

1

1.2

1.4

1.6

0.3

0.5

0.7

0.9

1.1

1.3

Cal

c.-O

bs. (

K) AIRS 80 (15 μm) AIRS 264 (4 μm)

AMSU 6

AIRS Radiative biases: Monthly evolution (5 years over the tropics)

“CO2” slopes (mean over 5 years):AIRS 80: 2.16 ppmv.yr-1

AIRS 264: 2.39 ppmv.yr-1

Mauna Loa: 2.05 ppmv.yr-1

~60 situations/month over sea~30 situations/month over land

03 04 05 06 07 08-0.2

0

0.2

0.4

0.6

0.8

03 04 05 06 07 08

03 04 05 06 07 08

0.6

0.8

1

1.2

1.4

1.6

0.3

0.5

0.7

0.9

1.1

1.3

Cal

c.-O

bs. (

K) AIRS 80 (15 μm) AIRS 264 (4 μm)

AMSU 6

A stand-alone approach

General features of the CO2 retrieval scheme :non-linear regressions

qCO2

Training of Neural Networks

Non-linear inference scheme

calc-obsbias removal

« clear sky »detection

Off-line

•Simultaneous use of IR and MW channels to decorrelate T/CO2.

IASI AMSU

•Retrieval limited to the tropical region.

Training data set (TIGR)

Selection of a set of CO2channels

[Chédin et al., JGR, 2003; Crevoisier et al., GRL, 2004]

Evaluation of the inference scheme characteristics

Pre

ssur

e (h

Pa)

We retrieve a mid-to-upper tropospheric integrated content of CO2.

5-15 km

10

100

1000

Mean CO2 averaging kernel over TIGR atmospheric dataset for nadir observation

AIRS/AMSUNOAA15IASI/AMSU

CO2 distribution from AIRS and IASI - Monthly average

Evaluation of IASI CO2

•Aircraft [Matsueda et al.]-8-10 km

•IASI CO2-integrated content 5-15 km

-1-2 points/month-until March. 2007

-period: July 2007-March 2008-monthly mean

JAL commercial airliners between Australia and Japan

10

100

1000

IASI CO2 weighting function

altitude of the aircraft

Seasonal cycle

Detrended CO2 seasonal cycle as observed in situ by JAL aircraft for 2003-2006

CO

2(p

pmv)

Apr Jun Aug Oct Dec Feb

-2

-1

0

1

2

Seasonal cycle

Apr Jun Aug Dec Feb

-2

-1

0

1

2

Detrended CO2 seasonal cycle as observed in situ by JAL aircraft for 2003-2006

CO

2(p

pmv)

Oct

IASI(07/07-08/03)

Latitudinal variationCO2 latitudinal variation in July

2006

2003

20052004

1 ppmv

2007

JAL aircraft (10 km)

IASI (5-15 km)

30N

-25N

25N

-20N

20N

-15N

15N

-10N

10N

-05N

05N

-EQ

EQ

-05S

05S

-10S

10S

-15S

15S

-20S

20S

-25S

Conclusions

The full information contained in the channels is needed (that excludes using PCA-like data).

•We retrieve a mid-to-upper tropospheric integrated content of CO2 from simultaneous IR/MW observations (TOVS, ATOVS, AIRS/AMSU, IASI/AMSU).

•The CO2 signal is very low:

-Reducing radiometric noise is as important as improving spectral resolution.

-A “good” AMSU instrument is important.

•Good agreement of CO2 distribution between IASI and AIRS but lower variability/uncertainty with IASI:

•General good agreement with in-situ observation in terms of seasonal cycle and latitudinal gradients.